Products

본문 바로가기
Products

Products

Products

Processor (NMP) family is shattering the barriers to deploying ML by delivering a general-purpose architecture and simple programmer’s model to enable virtually any class of neural network architecture and use case.

Our unique differentiation starts with the ability to simultaneously execute multiple AI/ML models significantly expanding the realm of capability over existing approaches. This game-changing advantage is provided by the co-developed NeuroMosAIc Studio software’s ability to dynamically allocate HW resources to match the target workload resulting in highly optimized, low-power execution. The designer may also select the optional on-device training acceleration extension enabling iterative learning post-deployment. This key capability cuts the cord to cloud dependence while elevating the accuracy, efficiency, customization, and personalization without reliance on costly model retraining and deployment, thereby extending device lifecycles.

The NeuroMosAIc Processor Family

Openly licensable silicon IP
portable to any process technology

NMP-300 – Lowest Power and Cost End Point Accelerator
  • Markets

    · Automotive
    · AIoT/Sensors
    · Industry 4.0
    · Smart Appliances
    · Wearables

  • Applications

    · Driver Authentication, Digital Mirrors, and Personalization
    · Predictive Maintenance
    · Machine Automation
    · Health Monitoring

  • Applications Example

  • Block-Diagram

  • Applications

    Performance Up to 512 GOPs
    MACs (8x8) 64, 128, 256
    Data Types 1-bit, INT8, INT16
    Internal SRAM Up to 512 KB
    Frameworks
NMP-500 – Performance Efficiency Accelerator
  • Markets

    · Automotive
    · Mobile and AR/VR
    · Drones and Robotics
    · Security and Surveillance
    · Medical Devices

  • Applications

    · Driver Monitoring and Fleet Management
    · Image / Video Analytics and Super Resolution
    · Intruder, Safety, and Compliance

  • Applications Example

  • Block-Diagram

  • Applications

    Performance Up to 4 TOPs
    MACs (8x8) 64, 128, 256
    Data Types 1-bit, INT8, INT16
    Internal SRAM Up to 4 MB
    Frameworks
NMP-700 – Performance Accelerator for Edge Computing
  • Markets

    · Automotive, AMR, and UAVs
    · AR/VR
    · Smart Building/Factory/City
    · Security and Surveillance
    · Telecommunications

  • Applications

    · Mobility and Autonomous Control
    · Process, Building, and Factory
    · Automation
    · Multi-Camera Stream Processing
    · Spectral Efficiency and Energy Mgmt

  • Applications Example

  • Block-Diagram

  • Applications

    Performance Up to 16 TOPs
    MACs (8x8) 4K, 8K
    Data Types 1-bit, INT8, INT16
    Internal SRAM Up to 16 MB
    Frameworks
NeuroMosAIc Studio
Optimize ML graphs produced from standard online training frameworks to target the chosen processor. The NeuroMosAIc Studio will convert, map, compile, simulate, and profile the model code through an intuitive series of steps allowing the user to maximize utilization and further optimize to application objectives. Once the objectives are met, the tools produce the runtime library in C/C++ and binary enabling easy integration with familiar embedded development toolkits.

NeuroMosAIc Studio

  • Converter

    Quantization
    (1bit / INT8/ INT16)

    Network Compression Network Optimization

    Network for NMP Version

    Precision Analysis

    Precision Adjustment

  • Mapper

    H/W level Optimization

    C code generation

  • Tools

    NMP Compiler

    NMP Simulator

  • Runtime Libraries

    Optimized A.I. functions

    Edge training

  • NMP Aware Training

    Optimize at training stage

    Optimized quantization

    aware training


logo

AiM Future Inc | #813, 114 Taebong-ro, Seocho-gu, Seoul, Republic of Korea | E-mail: info@aimfuture.ai

Copyright © AiM Future Inc. All rights reserved. Hosting by Whalessoft.